Creating a COVID-19 exposure alert app based on the Google exposure alert API - four considerations for public health agencies

The introduction of the Google/Apple exposure notification API means that public health agencies have a new tool available to help mitigate the risk associated with increased in-state and out-of-state travel. Creating an app based on this API requires a four pronged approach - each with its own challenges.

1. Creating the app itself.
The Google/Apple exposure notification isn't a stand alone app, rather it acts as the backbone for three critical issues:
a. tracking day to day exposures person to person
b. an alert mechanism to inform people when they have been exposed to someone who has tested positive
c. a method for an individual to broadcast their positive test results in an authenticated and anonymous fashion.

The code base provided by google on github acts as a quick start guide that can be deployed quickly into a virtual android phone to get a feel for the user experience and content needs. After spinning up the sample code you are presented with a series of pre-made screens and a basic U/X showing how a user might interact with the app.

The first thing you will notice is that there is a large amount of content that is needed in terms of how to use the app, what to do in the event of exposure, etc. This content needs to be state specific since there is no universal standard at the national level.

2. Developing a mechanism for testing notification.
In order to prevent malicious users from simply pinging the system with false positive test results, the app encourages the use of a QR code that would be unique to a test result. By scanning this QR code, the individual would verify their test result and upload that result into the server to trigger the notification process. This means that the local health agency must involve the testing team and epidemiologists to integrate the notification system with the testing process.

3. Developing the in-app content
There are considerable content needs within the app - all of which needs to be state specific. Not only is the copy state specific but in these ever changing times, the content needs to be timely and regularly updated. Involving the health communications team is critical as it having a plan for app updates to deploy the latest content.

4. App update and use
Without substantial uptake among the population the exposure notification app will not have a significant impact. During the planning and budgeting phase it is critical that adequate time and resources are made available to support the app. In this era of rampant disinformation and substantial distrust of public health officials, getting 'ahead of the curve' with respect to issues of privacy and data use will be critical. Starting this communication effort well before the app is available can help set the stage for uptake.

In combination with the pre-launch communications there will also be required post-launch communication in terms of social media response, public support and  question/answer periods.

While I believe that the exposure notification api and mobile applications present a tremendous opportunity, success will require a comprehensive approach that goes well beyond simply building a functional mobile app.

About the author:
Bill Patton has been working as a communication consultant with the Vermont Department of Health for the past 10 years and is currently working with the VDH team on COVID-19 communication projects.


Is social distancing and isolation working to flatten the curve of new corona virus infections? A quantitative approach

Is social distancing and social isolation working to flatten the curve of new corona virus infections and how do you measure the impact?

Here in my home state of Vermont I took the published data of positive test results to date (3/21) and then fit a "best fit" line to the existing data using google sheet formula GROWTH. I then compared the 'best fit line' against actual data to date to verify that the function was closely approximating current data.

I then projected this data out 7 days into the future. The resulting chart for Vermont looks like this

The actual number of positive test results on 3/21 was 49. The projected curve goes out 7 days into the future.

As shown the projected number of positives in 7 days is predicted to be 498.

Over the next 7 days I can plot the actual number as they are calculated and compare them against the prediction. If we "flatten the curve" the actual will be below the line.

Updated 3/22 - first sign of 'flatten the curve'